計(jì)量經(jīng)濟(jì)學(xué)實(shí)驗(yàn)報(bào)告

1、計(jì)量經(jīng)濟(jì)學(xué)實(shí)驗(yàn)報(bào)告 班級(jí)：2012級(jí)金融6班 姓名：趙勇志 學(xué)號(hào)：20125392實(shí)驗(yàn)一實(shí)驗(yàn)?zāi)康模赫莆找辉€性回歸模型的估計(jì)方法。實(shí)驗(yàn)要求：選擇方程進(jìn)行一元線性回歸。實(shí)驗(yàn)原理：普通最小二乘法（OLS）實(shí)驗(yàn)數(shù)據(jù)：東莞市經(jīng)濟(jì)部分?jǐn)?shù)據(jù)、廣東省宏觀經(jīng)濟(jì)部分?jǐn)?shù)據(jù)。 實(shí)驗(yàn)結(jié)果：1、REV作為應(yīng)變量，把GDP作為解釋變量Dependent Variable: REVMethod: Least SquaresDate: 06/08/14 Time: 14:06Sample: 1978 1995Included observations: 18VariableCoefficientStd. Errort-Sta

2、tisticProb. GDP0.0847810.00331125.604530.0000C-5826.1582517.475-2.3142860.0343R-squared0.976176 Mean dependent var38637.72Adjusted R-squared0.974687 S.D. dependent var48603.38S.E. of regression7732.823 Akaike info criterion20.84877Sum squared resid9.57E+08 Schwarz criterion20.94771Log likelihood-185

3、.6390 F-statistic655.5922Durbin-Watson stat0.335513 Prob(F-statistic)0.000000得到了估計(jì)方程：REV = 0.08478103497*GDP - 5826.1578622、把EXB作為應(yīng)變量，把REV作為解釋變量Dependent Variable: EXBMethod: Least SquaresDate: 06/08/14 Time: 14:09Sample: 1978 1995Included observations: 18VariableCoefficientStd. Errort-StatisticProb

4、. REV0.7193080.01115364.497070.0000C-2457.310680.5738-3.6106440.0023R-squared0.996168 Mean dependent var25335.11Adjusted R-squared0.995929 S.D. dependent var35027.97S.E. of regression2234.939 Akaike info criterion18.36626Sum squared resid79919268 Schwarz criterion18.46519Log likelihood-163.2963 F-st

5、atistic4159.872Durbin-Watson stat2.181183 Prob(F-statistic)0.000000得到了估計(jì)方程：EXB = 0.719308*REV - 2457.301把SLC作為應(yīng)變量，把GDP作為解釋變量Dependent Variable: SLCMethod: Least SquaresDate: 06/08/14 Time: 14:14Sample: 1978 1995Included observations: 18VariableCoefficientStd. Errort-StatisticProb. GDP0.4318270.00404

6、6106.72670.0000C-2411.3613076.237-0.7838670.4446R-squared0.998597 Mean dependent var224062.6Adjusted R-squared0.998510 S.D. dependent var244763.3S.E. of regression9449.149 Akaike info criterion21.24968Sum squared resid1.43E+09 Schwarz criterion21.34861Log likelihood-189.2471 F-statistic11390.59Durbi

7、n-Watson stat1.715091 Prob(F-statistic)0.000000得到了估計(jì)方程：SLC = 0.4318268605*GDP - 2411.36095LB作為應(yīng)變量，GDP1作為解釋變量Dependent Variable: LBMethod: Least SquaresDate: 06/08/14 Time: 14:19Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.4965380.00527794.102520.0000C

8、57.1378922.175202.5766570.0176R-squared0.997634 Mean dependent var1486.166Adjusted R-squared0.997521 S.D. dependent var1556.667S.E. of regression77.49819 Akaike info criterion11.62133Sum squared resid126125.4 Schwarz criterion11.72007Log likelihood-131.6453 F-statistic8855.284Durbin-Watson stat1.781

9、264 Prob(F-statistic)0.000000得到了估計(jì)方程：LB = 0.4965380887*GDP1 + 57.13788724ZJ作為應(yīng)變量，GDP1作為解釋變量Dependent Variable: ZJMethod: Least SquaresDate: 06/08/14 Time: 14:23Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1654060.00294056.260930.0000C-42.9895112.35547

10、-3.4793900.0022R-squared0.993409 Mean dependent var433.0448Adjusted R-squared0.993095 S.D. dependent var519.6546S.E. of regression43.18007 Akaike info criterion10.45158Sum squared resid39154.89 Schwarz criterion10.55032Log likelihood-118.1931 F-statistic3165.292Durbin-Watson stat0.405988 Prob(F-stat

11、istic)0.000000 得到了估計(jì)方程：ZJ = 0.1654055246*GDP1 - 42.9895126SE作為應(yīng)變量，GDP1作為解釋變量Dependent Variable: SEMethod: Least SquaresDate: 06/08/14 Time: 14:29Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1664470.00345948.124650.0000C-36.6064514.53536-2.5184410.0200

12、R-squared0.991014 Mean dependent var442.4265Adjusted R-squared0.990586 S.D. dependent var523.5596S.E. of regression50.79837 Akaike info criterion10.77655Sum squared resid54189.96 Schwarz criterion10.87529Log likelihood-121.9303 F-statistic2315.982Durbin-Watson stat0.633700 Prob(F-statistic)0.000000得

13、到了估計(jì)方程：SE = 0.1664474631*GDP1 - 36.60645462YY作為應(yīng)變量，GDP1作為解釋變量Dependent Variable: YYMethod: Least SquaresDate: 06/08/14 Time: 14:31Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1716080.00616127.855130.0000C22.4615625.891060.8675410.3955R-squared0.973648

14、 Mean dependent var516.3478Adjusted R-squared0.972393 S.D. dependent var544.5861S.E. of regression90.48441 Akaike info criterion11.93117Sum squared resid171936.0 Schwarz criterion12.02991Log likelihood-135.2085 F-statistic775.9083Durbin-Watson stat1.099129 Prob(F-statistic)0.000000由于常數(shù)項(xiàng)沒有通過檢驗(yàn)，所以去掉常數(shù)

15、項(xiàng)重新檢驗(yàn)。Dependent Variable: YYMethod: Least SquaresDate: 06/08/14 Time: 14:38Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1752690.00446439.261420.0000R-squared0.972704 Mean dependent var516.3478Adjusted R-squared0.972704 S.D. dependent var544.5861S.E. o

16、f regression89.97426 Akaike info criterion11.87943Sum squared resid178098.1 Schwarz criterion11.92880Log likelihood-135.6134 Durbin-Watson stat1.058314 得到了估計(jì)方程：YY = 0.1752685914*GDP1CS作為應(yīng)變量，SE作為解釋變量Dependent Variable: CSMethod: Least SquaresDate: 06/08/14 Time: 14:43Sample: 1978 2000Included observa

17、tions: 23VariableCoefficientStd. Errort-StatisticProb. SE0.4905460.01350936.311580.0000C29.031219.1419293.1756120.0046R-squared0.984323 Mean dependent var246.0617Adjusted R-squared0.983576 S.D. dependent var258.8672S.E. of regression33.17508 Akaike info criterion9.924416Sum squared resid23112.31 Sch

18、warz criterion10.02316Log likelihood-112.1308 F-statistic1318.531Durbin-Watson stat1.281007 Prob(F-statistic)0.000000 得到了估計(jì)方程：CS = 0.4905459191*SE + 29.03121437CZ作為應(yīng)變量，CS作為解釋變量Dependent Variable: CZMethod: Least SquaresDate: 06/08/14 Time: 14:49Sample: 1978 2000Included observations: 23VariableCoeff

19、icientStd. Errort-StatisticProb. CS1.2620160.02737746.098370.0000C-20.247229.665329-2.0948300.0485R-squared0.990215 Mean dependent var290.2865Adjusted R-squared0.989749 S.D. dependent var328.3047S.E. of regression33.24050 Akaike info criterion9.928356Sum squared resid23203.55 Schwarz criterion10.027

20、09Log likelihood-112.1761 F-statistic2125.060Durbin-Watson stat1.475902 Prob(F-statistic)0.000000得到了估計(jì)方程：CZ = 1.26201555*CS - 20.24721934實(shí)驗(yàn)二實(shí)驗(yàn)?zāi)康模赫莆找辉€性回歸模型的檢驗(yàn)方法。實(shí)驗(yàn)要求：進(jìn)行經(jīng)濟(jì)、擬合優(yōu)度、參數(shù)顯著性和方程顯著性等檢驗(yàn)。實(shí)驗(yàn)原理：擬合優(yōu)度的判定系數(shù)R2 檢驗(yàn)和參數(shù)顯著性t檢驗(yàn)等。實(shí)驗(yàn)結(jié)果：1.把REV作為應(yīng)變量，把GDP作為解釋變量REV = 0.08478103497*GDP - 5826.157862 （0.003311） （25

21、17.475） （25.60453） （-2.314286）R2= 0.976176 SE = 7732.823REV對(duì)GDP的回歸系數(shù)為 0.08478103497，R2= 0.976176 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量GDP在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。2.把EXB作為應(yīng)變量，把REV作為解釋變量EXB = 0.7193079524*REV - 2457.309747 (0.011153) (0.011153) (64.49707) ( -3.610644)R2 = 0.996168 SE = 2234.939EXB對(duì)R

22、EV的回歸系數(shù)為 0.7193079524，R2= 0.996168 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量REV在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。3.把SLC作為應(yīng)變量，把GDP作為解釋變量SLC = 0.4318268605*GDP - 2411.36095(0.004046) (3076.237)(106.7267) (-0.783867)R2= 0.998597 SE= 9449.149SLC對(duì)GDP的回歸系數(shù)為0.4318268605 ，R2= 0.998597 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16

23、)，說明解釋變量GDP在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。4.把LB作為應(yīng)變量，GDP1作為解釋變量LB = 0.4965380887*GDP1 + 57.13788724(0.005277) (22.17520)(94.10252) (2.576657)R2=0.997634 SE= 77.49819LB對(duì)GDP1的回歸系數(shù)為0.4965380887 ，R2= 0.997634 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量GDP1在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。5.把ZJ作為應(yīng)變量，GDP1作為解釋變量ZJ = 0.1654055

24、246*GDP1 - 42.9895126(0.002940) (12.35547)(56.26093) (-3.479390)R2= 0.993409 SE = 43.18007ZJ對(duì)GDP1的回歸系數(shù)為 0.1654055246，R2=0.993409 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量GDP1在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。6.把SE作為應(yīng)變量，GDP1作為解釋變量SE = 0.1664474631*GDP1 - 36.60645462(0.003459) (14.53536)(48.12465) (-2.518441)R2=

25、 0.991014 SE = 50.79837SE對(duì)GDP1的回歸系數(shù)為 0.1664474631，R2= 0.991014，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量GDP1在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。7.把YY作為應(yīng)變量，GDP1作為解釋變量YY = 0.1752685914*GDP1 (0.004464) (39.26142) R2= 0.972704 SE = 89.97426YY對(duì)GDP1的回歸系數(shù)為0.1752685914 ，R2= 0.972704 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解

26、釋變量GDP1在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。8. CS作為應(yīng)變量，SE作為解釋變量 CS = 0.4905459191*SE + 29.03121437(0.013509) (9.141929) (36.31158) (3.175612)R2= 0.984323 SE = 33.17508CS對(duì)SE的回歸系數(shù)為0.4905459191 ，R2= 0.984323 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量SE在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。9.CZ作為應(yīng)變量，CS作為解釋變量 CZ = 1.26201555*CS - 20.

27、24721934 (0.027377) (9.665329)(46.09837) (-2.094830)R2= 0.990215 SE = 33.24050CZ對(duì)CS的回歸系數(shù)為 1.26201555，R2= 0.990215 ，接近1，因此擬合優(yōu)度好。t（16）=2.12，l t lt(16)，說明解釋變量CS在95%的置信度下顯著，即通過變量顯著性檢驗(yàn)。實(shí)驗(yàn)三 實(shí)驗(yàn)?zāi)康模赫莆斩嘣€性回歸模型的估計(jì)和檢驗(yàn)方法。實(shí)驗(yàn)要求：選擇方程進(jìn)行多元線性回歸。實(shí)驗(yàn)原理：普通最小二乘法（OLS）。實(shí)驗(yàn)結(jié)果：根據(jù)東莞數(shù)據(jù)選擇第二產(chǎn)業(yè)增加值（GDP2）、固定資產(chǎn)凈值（NKF2）和勞動(dòng)者人數(shù)（LT2）的數(shù)據(jù)，把G

28、DP2作為應(yīng)變量，NKF2和LT2作為兩個(gè)解釋變量進(jìn)行二元線性回歸分析。 1、 作GDP2與NKF2、GDP2與LT2的散點(diǎn)圖Scat GDP2 NKF2Scat GDP2 LT2Dependent Variable: GDP2Method: Least SquaresDate: 06/15/14 Time: 14:12Sample: 1978 1995Included observations: 18VariableCoefficientStd. Errort-StatisticProb. NKF20.6982960.02151232.461030.0000C55714.2411197.58

29、4.9755620.0001R-squared0.985043 Mean dependent var264711.6Adjusted R-squared0.984108 S.D. dependent var308327.3S.E. of regression38868.85 Akaike info criterion24.07821Sum squared resid2.42E+10 Schwarz criterion24.17714Log likelihood-214.7039 F-statistic1053.718Durbin-Watson stat1.306346 Prob(F-stati

30、stic)0.000000得到了估計(jì)方程：GDP2 = 0.6982962506*NKF2 + 55714.24213Dependent Variable: GDP2Method: Least SquaresDate: 06/15/14 Time: 14:18Sample: 1978 1995Included observations: 18VariableCoefficientStd. Errort-StatisticProb. LT22.7109800.4400036.1612740.0000C-431249.1120096.7-3.5908500.0024R-squared0.70349

31、1 Mean dependent var264711.6Adjusted R-squared0.684959 S.D. dependent var308327.3S.E. of regression173059.4 Akaike info criterion27.06510Sum squared resid4.79E+11 Schwarz criterion27.16403Log likelihood-241.5859 F-statistic37.96130Durbin-Watson stat0.325257 Prob(F-statistic)0.000014得到了估計(jì)方程：GDP2 = 2.

32、71097983*LT2 - 431249.0779作GDP2與 NKF2和LT2的二元回歸Dependent Variable: GDP2Method: Least SquaresDate: 06/15/14 Time: 14:23Sample: 1978 1995Included observations: 18VariableCoefficientStd. Errort-StatisticProb. NKF20.6293780.02971521.180110.0000LT20.3953140.1365112.8958380.0111C-25143.3329418.08-0.8546900

33、.4062R-squared0.990406 Mean dependent var264711.6Adjusted R-squared0.989127 S.D. dependent var308327.3S.E. of regression32150.29 Akaike info criterion23.74524Sum squared resid1.55E+10 Schwarz criterion23.89364Log likelihood-210.7072 F-statistic774.2594Durbin-Watson stat1.983844 Prob(F-statistic)0.00

34、0000得到了估計(jì)方程：GDP2 = 0.6293775361*NKF2 + 0.3953138872*LT2 - 25143.332112、作LB與GDP1的一元回歸Dependent Variable: LBMethod: Least SquaresDate: 06/15/14 Time: 14:32Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.5058490.004310117.37030.0000R-squared0.996886 Mean de

35、pendent var1486.166Adjusted R-squared0.996886 S.D. dependent var1556.667S.E. of regression86.86456 Akaike info criterion11.80908Sum squared resid165999.9 Schwarz criterion11.85845Log likelihood-134.8044 Durbin-Watson stat1.395510作LB與 GDP1、T的二元回歸；Dependent Variable: LBMethod: Least SquaresDate: 06/15

36、/14 Time: 14:39Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.4787580.00937651.063660.0000T9.0105162.8737003.1355110.0050R-squared0.997879 Mean dependent var1486.166Adjusted R-squared0.997778 S.D. dependent var1556.667S.E. of regression73.37653 Akaike i

37、nfo criterion11.51203Sum squared resid113066.4 Schwarz criterion11.61077Log likelihood-130.3883 Durbin-Watson stat1.917391估計(jì)方程的判定系數(shù)R2 分別接近1，參數(shù)顯著性t檢驗(yàn)值均大于2，方程顯著性F檢驗(yàn)顯著。調(diào)整的判定系數(shù)為0.997778 ，比一元回歸有明顯改善。所以，得到了估計(jì)方程：LB = 0.4787579113*GDP1 + 9.010515807*TZJ與 GDP1的一元回歸Dependent Variable: ZJMethod: Least SquaresD

38、ate: 06/15/14 Time: 14:43Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1584000.00262860.271220.0000R-squared0.989610 Mean dependent var433.0448Adjusted R-squared0.989610 S.D. dependent var519.6546S.E. of regression52.96956 Akaike info criterion10.81982

39、Sum squared resid61727.04 Schwarz criterion10.86919Log likelihood-123.4279 Durbin-Watson stat0.278503作ZJ與 GDP1、T的二元回歸；Dependent Variable: ZJMethod: Least SquaresDate: 06/15/14 Time: 14:53Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1811640.00435141.63

40、5220.0000T-7.5713031.333671-5.6770410.0000R-squared0.995901 Mean dependent var433.0448Adjusted R-squared0.995706 S.D. dependent var519.6546S.E. of regression34.05371 Akaike info criterion9.976696Sum squared resid24352.76 Schwarz criterion10.07543Log likelihood-112.7320 Durbin-Watson stat0.637365估計(jì)方程

41、的判定系數(shù)R2 分別接近1，參數(shù)顯著性t檢驗(yàn)值均大于2，方程顯著性F檢驗(yàn)顯著。調(diào)整的判定系數(shù)為0.995706 ，比一元回歸有明顯改善。所以，得到了估計(jì)方程：ZJ = 0.1811640816*GDP1 - 7.571303452*T作SE與GDP1的一元回歸Dependent Variable: SEMethod: Least SquaresDate: 06/15/14 Time: 14:57Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1604820

42、.00281057.114940.0000R-squared0.988300 Mean dependent var442.4265Adjusted R-squared0.988300 S.D. dependent var523.5596S.E. of regression56.63148 Akaike info criterion10.95351Sum squared resid70556.74 Schwarz criterion11.00288Log likelihood-124.9654 Durbin-Watson stat0.523106作SE 與 GDP1、T的二元回歸；Depende

43、nt Variable: SEMethod: Least SquaresDate: 06/15/14 Time: 15:01Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1806280.00566531.882960.0000T-6.7006471.736460-3.8587970.0009R-squared0.993154 Mean dependent var442.4265Adjusted R-squared0.992828 S.D. depende

44、nt var523.5596S.E. of regression44.33846 Akaike info criterion10.50452Sum squared resid41283.89 Schwarz criterion10.60326Log likelihood-118.8020 Durbin-Watson stat0.763060估計(jì)方程的判定系數(shù)R2 分別接近1，參數(shù)顯著性t檢驗(yàn)值均大于2，方程顯著性F檢驗(yàn)顯著。調(diào)整的判定系數(shù)為0.992828 ，比一元回歸有明顯改善。所以，得到了估計(jì)方程：SE = 0.1806284457*GDP1 - 6.70064709*T作YY與 GDP1

45、的一元回歸Dependent Variable: YYMethod: Least SquaresDate: 06/15/14 Time: 15:15Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. GDP10.1752690.00446439.261420.0000R-squared0.972704 Mean dependent var516.3478Adjusted R-squared0.972704 S.D. dependent var544.5861S.E. of

46、 regression89.97426 Akaike info criterion11.87943Sum squared resid178098.1 Schwarz criterion11.92880Log likelihood-135.6134 Durbin-Watson stat1.058314作YY與 GDP1、T的二元回歸Dependent Variable: YYMethod: Least SquaresDate: 06/15/14 Time: 15:25Sample: 1978 2000Included observations: 23VariableCoefficientStd.

47、 Errort-StatisticProb. GDP10.1594490.01115514.294190.0000T5.2618073.4189951.5389920.1387R-squared0.975470 Mean dependent var516.3478Adjusted R-squared0.974302 S.D. dependent var544.5861S.E. of regression87.29999 Akaike info criterion11.85952Sum squared resid160047.1 Schwarz criterion11.95826Log like

48、lihood-134.3845 Durbin-Watson stat1.198126估計(jì)方程的判定系數(shù)R2 分別接近1，參數(shù)t檢驗(yàn)GDP1顯著，方程顯著性F檢驗(yàn)顯著。調(diào)整的判定系數(shù)為0.974302 ，比一元回歸有明顯改善。所以，得到了估計(jì)方程：YY = 0.1594485548*GDP1 + 5.261806944*T實(shí)驗(yàn)四第一部分 異方差模型的檢驗(yàn)實(shí)驗(yàn)?zāi)康模赫莆债惙讲钅Ｐ偷臋z驗(yàn)方法。實(shí)驗(yàn)要求：掌握?qǐng)D形法檢驗(yàn)和Glejser檢驗(yàn)。實(shí)驗(yàn)原理：圖形法檢驗(yàn)、Glejser檢驗(yàn)。實(shí)驗(yàn)結(jié)果：1、ZJ對(duì)GDP1和T回歸的殘差趨勢(shì)圖和殘差散點(diǎn)圖。genr e2=resid2從圖上看ZJ對(duì)GDP1和T回歸的

49、殘差存在異方差。2、做對(duì)ZJ和GDP1回歸的Glejser檢驗(yàn)。genr E1=residls abs(e1) c gdp1ls abs(e1) c gdp12ls abs(e1) c SQR(gdp1)ls abs(e1) c 1/gdp1以上四個(gè)對(duì)gdp1回歸得結(jié)果分別為：對(duì)gdp1Dependent Variable: ABS(E1)Method: Least SquaresDate: 06/16/14 Time: 17:21Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-Statistic

50、Prob. C20.718966.0638983.4167730.0026GDP10.0043870.0014433.0403050.0062R-squared0.305635 Mean dependent var33.34424Adjusted R-squared0.272570 S.D. dependent var24.84736S.E. of regression21.19219 Akaike info criterion9.028084Sum squared resid9431.287 Schwarz criterion9.126822Log likelihood-101.8230 F

51、-statistic9.243454Durbin-Watson stat1.148496 Prob(F-statistic)0.006221對(duì)gdp1 2回歸得結(jié)果為：Dependent Variable: ABS(E1)Method: Least SquaresDate: 06/16/14 Time: 17:24Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. C24.836605.3297584.6599850.0001GDP124.82E-071.64E-072.

52、9285480.0080R-squared0.289974 Mean dependent var33.34424Adjusted R-squared0.256164 S.D. dependent var24.84736S.E. of regression21.42984 Akaike info criterion9.050387Sum squared resid9643.999 Schwarz criterion9.149126Log likelihood-102.0795 F-statistic8.576393Durbin-Watson stat1.132628 Prob(F-statist

53、ic)0.008028從F檢驗(yàn)來為來看整個(gè)模型不顯著(3)對(duì)SQR(gdp1)回歸得結(jié)果為：Dependent Variable: ABS(E1)Method: Least SquaresDate: 06/16/14 Time: 17:29Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. C12.165158.5006121.4310910.1671SQR(GDP1)0.4640730.1584552.9287360.0080R-squared0.290001 Mean

54、 dependent var33.34424Adjusted R-squared0.256191 S.D. dependent var24.84736S.E. of regression21.42944 Akaike info criterion9.050350Sum squared resid9643.639 Schwarz criterion9.149088Log likelihood-102.0790 F-statistic8.577497Durbin-Watson stat1.118705 Prob(F-statistic)0.008024常數(shù)項(xiàng)不顯著，去掉常數(shù)項(xiàng)再進(jìn)行回歸得結(jié)果為：D

55、ependent Variable: ABS(E1)Method: Least SquaresDate: 06/16/14 Time: 17:31Sample: 1978 2000Included observations: 23VariableCoefficientStd. Errort-StatisticProb. SQR(GDP1)0.6569810.0852537.7062760.0000R-squared0.220758 Mean dependent var33.34424Adjusted R-squared0.220758 S.D. dependent var24.84736S.E

56、. of regression21.93392 Akaike info criterion9.056451Sum squared resid10584.13 Schwarz criterion9.105820Log likelihood-103.1492 Durbin-Watson stat1.019181(4)對(duì) 1/gdp1回歸得結(jié)果為：Dependent Variable: ABS(E1)Method: Least SquaresDate: 06/16/14 Time: 17:36Sample: 1978 2000Included observations: 23VariableCoef

57、ficientStd. Errort-StatisticProb. C39.308827.0213465.5984740.00001/GDP1-3994.3633218.612-1.2410200.2283R-squared0.068328 Mean dependent var33.34424Adjusted R-squared0.023963 S.D. dependent var24.84736S.E. of regression24.54784 Akaike info criterion9.322066Sum squared resid12654.53 Schwarz criterion9

58、.420805Log likelihood-105.2038 F-statistic1.540131Durbin-Watson stat0.869730 Prob(F-statistic)0.228283從F檢驗(yàn)來為來看整個(gè)模型不顯著 從四個(gè)回歸的結(jié)果看，回歸(4)不顯著，(1)、（2）、(3)顯著，比較(1)(3)不帶常數(shù)項(xiàng)的回歸,選擇（3），方程為：ABS(E1) = 0.6569810506*SQR(GDP1)即異方差的形式為：2=( 0.6569810506*SQR(GDP1)2=0.4316241008*也即異方差的形式為：2=2*第二部分 異方差模型的處理實(shí)驗(yàn)?zāi)康模赫莆债惙讲钅Ｐ偷?/p>

59、處理方法。實(shí)驗(yàn)要求：理解同方差性變換，掌握加權(quán)最小二乘法（WLS）。實(shí)驗(yàn)原理：加權(quán)最小二乘法（WLS）、同方差性變換和廣義最小二乘法（GLS）。實(shí)驗(yàn)結(jié)果： 1、 已知ZJ對(duì)GDP1和T回歸異方差的形式為： 把作為權(quán)數(shù)來進(jìn)行加權(quán)最小二乘法。得到回歸結(jié)果為：Dependent Variable: ZJMethod: Least SquaresDate: 06/22/14 Time: 14:28Sample: 1978 2000Included observations: 23Weighting series: 1/(GDP1)1/6VariableCoefficientStd. Errort-St

60、atisticProb. GDP10.1589580.00464634.210210.0000T-3.4691890.487355-7.1183960.0000Weighted StatisticsR-squared0.706997 Mean dependent var87.15467Adjusted R-squared0.693044 S.D. dependent var13.22542S.E. of regression7.327359 Akaike info criterion6.904049Sum squared resid1127.494 Schwarz criterion7.002